Longitudinal Study of Depressive Symptoms and Health-Related Quality of Life During Pregnancy and After Delivery: The Health Status in Pregnancy (HIP) Study
- First Online:
- Cite this article as:
- Setse, R., Grogan, R., Pham, L. et al. Matern Child Health J (2009) 13: 577. doi:10.1007/s10995-008-0392-7
- 435 Views
Objective Depressive symptoms are known to affect functioning in early pregnancy. We estimated the effect of a change in depressive symptoms status on health-related quality of life (HRQoL) throughout pregnancy and after delivery. Methods Longitudinal study of 200 women. The independent variable was depressive symptoms, defined as a Center for Epidemiologic Studies Depression (CES-D) score of ≥16. The dependent variable was HRQoL from 8 domains of the Medical Outcomes Study (SF-36) Short Form. Women were categorized based on the change in CES-D score: (1) never depressed, (2) became well, (3) became depressed and (4) always depressed. A random effects model was used to (1) estimate the effect of a change in depressive symptomatology from the first to the second trimester on HRQOL in the second trimester and (2) estimate the change in depressive symptomatology from the second to the third trimester on HRQoL in the third trimester and after delivery, adjusting for covariates. Intra-individual correlations were accounted for using generalized estimating equations (GEE). Results The proportion of women with depressive symptoms was 15%, 14%, and 30% in the first, second and third trimesters, respectively, and 9% after delivery. Women who became depressed had scores in the social domains that were 10–23 points and 19–31 points lower in the second and third trimesters, respectively, compared to women with no depressive symptoms. Women who became well had scores that were 3–31 points lower, compared to women with no depressive symptoms. Conclusions Alterations in depressive symptomatology have a substantial effect on functioning during pregnancy and after delivery.
KeywordsPregnancyDepressive symptomsHealth-related quality of lifeLongitudinal analysis
Pregnancy is a common event among reproductive-age women and is often thought to be a time of happiness for the expectant mother. The results of several investigations, however, suggest that depressive symptoms can be highly prevalent during pregnancy [1–3] and after delivery [4, 5], with potentially substantial effects on maternal and newborn outcomes [6–8]. Estimates of the prevalence rates of depressive symptoms among women of childbearing age range from 10% to 50%, depending on the instrument used and the demographic characteristics of the study population [9–11]. Zayas and colleagues  found that 50% of pregnant women in the third trimester had depressive symptomatology. Depressive symptoms during pregnancy have been linked to both preterm births and low birth weight infants . Postpartum depressive symptoms have been shown to adversely affect mother-infant interactions and cognitive development in the newborn [12, 13].
Perceived health-related quality of life has emerged as an important outcome in adults [14, 15]. In primary care patients with chronic medical conditions, depressive symptoms have been shown to further limit physical and mental functioning . Women with depressive symptoms in early pregnancy and the postpartum period are reported to have poorer functioning compared to women with no depressive symptoms [1–3]. In earlier work, we found that functional status scores among women with depressive symptoms in early pregnancy were 10 to 20 points lower compared to women without depressive symptoms . Howell and colleagues  reported substantially lower scores among women with depressive symptoms in the months following delivery.
While prior studies suggest diminished functioning in pregnant women with depressive symptoms [3, 5, 16], the influence of depressive symptomatology on maternal health status over the course of pregnancy has garnered relatively little attention from epidemiological studies [1, 2]. Much of the literature is cross-sectional in nature, with one measure of the prevalence of depressive symptomatology rather than multiple estimates of prevalence over the course of pregnancy and the puerperium. Also, prior studies have limited their analyses to specific functional domains [1, 2] rather than examining each of the eight dimensions of health-related quality of life.
Our objectives were to (1) estimate the prevalence of depressive symptomatology during pregnancy and the puerperium and (2) estimate the independent effect of depressive symptomatology on HRQoL during pregnancy and after delivery (within 3 days postpartum) through a longitudinal approach. Because hormonal fluctuations after delivery can be associated with symptoms of depression, we hypothesized that the prevalence of depressive symptomatology would be highest after delivery. We also hypothesized that depressive symptoms would be independently associated with poorer physical and mental functioning, after adjustment for participant characteristics and clinical complications. Estimating the prevalence of depressive symptoms can assist in the development of screening protocols for expectant mothers. Increasing attention to the effect of depressive symptomatology on maternal functioning may inform the development of guidelines to improve maternal well-being.
Research Design and Methods
We conducted a longitudinal study of 200 women receiving prenatal care in Baltimore City as part of the Health Status in Pregnancy (HIP) Study to estimate the effect of the change in depressive symptomatology on HRQoL using a random effects model. The HIP Study was a longitudinal cohort study of health-related quality of life in pregnant women in an urban setting. Women were eligible to participate in this study if they were: (1) 18 years old or older, (2) presented for antenatal care at either of the two outpatient settings, (3) intended to maintain their pregnancies and deliver within Baltimore city, (4) 14 weeks gestation or less at the time of recruitment, (5) able to provide written informed consent and (6) not diagnosed with HIV or any active cancers. This study was approved by the institutional review board of the Johns Hopkins School of Medicine.
Recruitment was conducted in two university-based outpatient practices between July 2004 and September 2005. Gestational age was determined through a review of the electronic medical record and was based on either the obstetrician’s assessment of the last menstrual period (LMP) or both the LMP and obstetric ultrasound assessment. If there was a discrepancy between the gestational age by LMP and ultrasound, the estimated gestational age determined by ultrasound was assigned to the participant.
We collected data on both depressive symptomatology and health-related quality of life at four time points. Assessments were made at 14 weeks (visit 1) at baseline, 18–22 weeks (visit 2), 28–32 weeks (visit 3) and within 3 days after delivery (visit 4). Data on clinical factors was collected from electronic prenatal records at baseline and at each followup visit.
Depressive symptomatology was measured using the Center for Epidemiologic Studies Depression (CES-D) Scale. The CES-D Scale is a 20 item self-administered questionnaire with well established reliability and validity in both the general population (Cronbach’s alpha = 0.84) and among postpartum women (0.88–0.91) [17, 18]. The questionnaire asks respondents about the presence of sadness, crying and hopelessness or changes in appetite or sleep. The scale has a sensitivity of 80–90% in primary care settings with a cutoff of 16 or more for clinical depression. Prior studies have used the CES-D in pregnancy with a moderate sensitivity of 80%  and a specificity of 98–99% . Items on the CES-D are rated on a zero-to–3 point response scale. The total score is determined by summing the ratings across all 20 items, with possible scores ranging between 0 and 60. A threshold of (16 is considered to be an indicator of clinically significant elevations in depressive symptomatology in general populations and among pregnant women . Forty to fifty percent of persons with scores (16 would be classified as clinically depressed . We administered the questionnaire at each of the four study visits.
Health-Related Quality of Life
Health-related quality of life (HRQoL) was measured using the Medical Outcomes Study Short Form-36 Health Questionnaire (SF-36), a multidimensional measure of health status designed for self or interviewer administration. This questionnaire is widely used in clinical research and has been established as a reliable and valid measure of health related quality of life in different populations . The questionnaire measures perceptions of health related quality of life in 8 domains, including Physical Functioning, Role-Physical, Bodily Pain, General Health, Vitality, Role-Emotional, Social Functioning and Mental Health. Physical Functioning assesses limitations in physical activities because of health problems. Role-physical assesses problems with daily activities as a result of physical health, while Role-Emotional assesses problems with activities as a result of emotional problems. Vitality measures perception of fatigue or energy. Bodily Pain measures the extent of pain. General Health measures perception of health based on physical limitations, emotional limitations and well-being. Social Functioning measures limitations in social activities because of physical or emotional problems. Mental Health measures psychological well-being and distress. Questionnaire responses are scored on a 5-point scale. These absolute scores are then transformed into a score of 0–100 with higher scores indicating better functioning or well being.
We collected data on socio-demographics at baseline from electronic patient records. These factors included maternal age, race (white, black, Asian, Hispanic, other), marital status (married, single), educational level, payment source (Medicaid, commercial insurance) employment status at time of survey (employed, unemployed) and total household income (<$15,000, $15,001–$35,000, $35,001–$50,000, >$50,000).
Clinical variables included smoking status (never, past, current) and first trimester body mass index (kg/m2). Current pregnancy-related medical conditions were categorized as hypertensive disease (chronic hypertension, transient hypertension), diabetes mellitus (gestational and type 2 diabetes), heart disease, sexually transmitted diseases (gonorrhea, chlamydia, syphilis, Hepatitis B), abnormal vaginal bleeding, abnormal cervical cytology requiring intervention, pyelonephritis, asthma, preterm labor, and diagnosed mental health disorders (depression, anxiety). Past medical conditions included chronic hypertension, diabetes mellitus, heart disease, sexually transmitted disease, infertility, renal disease, asthma, history of psychiatric diagnosis (depression, anxiety) and prior birth outcomes (term delivery, preterm delivery, fetal death). Also, we collected data on current birth outcomes (term delivery, preterm delivery, fetal death), method of delivery (cesarean, vaginal, vaginal-assisted delivery) and complications of delivery (pre-eclampsia, premature rupture of membranes, chorioamnionitis, postpartum hemorrhage, postpartum fever), newborn characteristics (newborn nursery versus neonatal intensive care unit) and infant birth weight (<2,500 grams versus (2,500 grams).
Earlier work has shown an association between social support and depressive symptoms . We assessed social support at baseline using a modified version of the Norbeck Social Support Questionnaire . Participants were asked the following questions: (1) “Do you get emotional support from your spouse/boyfriend/significant other?” (2) “How much of your support is provided by your spouse/boyfriend/significant other?” and (3) “Do you receive emotional support from people other than your significant other?”
Socio-demographic and clinical factors were compared between women with and without depressive symptomatology using t-test and chi-square analysis as appropriate. Based on an alpha of 0.05, there was 80% power to detect a 25% difference in socio-demographic characteristics between the two groups. There was over 80% power to detect a six point difference or higher in health-related quality of life scores between women with and without depressive symptomatology.
We estimated the adjusted proportion of women with depressive symptoms with 95% confidence intervals for the first, second and third trimesters and after delivery. Bivariate analysis was used to estimate the unadjusted relation of depressive symptoms with each HRQoL domain and with participant characteristics. We conducted a multivariate analysis to estimate the association of depressive symptoms with HRQoL in the first trimester.
We conducted a longitudinal analysis using a random effects model to (1) estimate the effect of the change in depressive (baseline to second trimester) symptomatology on HRQoL in the second trimester, (2) the change in depressive (second to third trimester) symptoms on HRQoL in the third trimester and (3) the change in depressive (second to third trimester) symptoms on HRQoL after delivery. We classified women’s depressive symptom status based on a change in the depressive symptomatology score from less than 16 to 16 or higher. Women were therefore categorized as: (1) never depressed (score below 16), (2) became depressed (score changed from less than 16 to 16 or higher), (3) became well (score changed from 16 or higher to less than 16) and (4) always depressed (score remained at 16 or higher). We estimated adjusted regression coefficients and 95% confidence intervals using multiple linear regression and random effects models [25, 26]. Intra-individual correlations were accounted for using generalized estimating equations (GEE) .
Variables for the multivariate models were selected on the basis of a priori hypotheses or bivariate associations. These variables included maternal age, race, payment source, marital status, employment status, income, body mass index, one or more current pregnancy-related complications, one or more prior medical conditions, gestational age and the level of social support from the significant other (little or moderate versus a great deal). The postpartum model was adjusted for the variables outlined above as well as for method of delivery, infant birth weight, one or more delivery complications and neonatal intensive care admission. Because functional status scores can be skewed, we conducted sensitivity analyses using log transformed functional scores and median scores. Because we found little difference in the results, we chose to present the regression coefficients from the linear regression models for ease of interpretation.
Characteristics of Study Participants
Of the 250 women who were eligible for the Health Status in Pregnancy Study, 200 (80%) women initially agreed to participate in the study. Response rates for the second, third and fourth study visits were 93%, 86%, and 81%, respectively. Women who did not complete all 4 surveys had similar demographic characteristics (age, race, payment source and parity) and functional status scores at baseline compared to the women who did complete each survey.
Characteristics of study sample at baseline (N = 200)
No depressive symptoms (CES-D < 16) (n = 170)
Depressive symptoms (CES-D ≥ 16) (n = 30)
Age (yrs), mean (SD)
Marital Status (%)
Employment Status (%)
Income level (%)
Support by significant othera
Amount of support by significant other
Little or moderate
A great deal
Most support other than significant other
One or more prior live births
Prior Preterm birth
Prior spontaneous abortion
One or more prior medical conditionsb (%)
One or more current pregnancy conditionsc
Body Mass Index (BMI), (kg/m2), mean (SD),
26.6 ± 7.8
29.2 ± 10.4
Gestational age at baseline, (weeks)
9.0 ± 3.5
8.4 ± 3.0
Method of delivery (%)
Vaginal or vaginal-assisted
Cesarean delivery (primary or repeat)
Live birth (%)
Birth weight <2,500 g (%)
NICU admission (%)
One or more delivery complications (%)d
Proportion of Participants with Depressive Symptoms
Depressive Symptoms and HRQoL in First Trimester
Change in Depressive Symptoms and HRQoL in the Second Trimester
Health-related quality of life in the second trimester by change in depressive symptomatology
Women without depressive symptoms in first or second trimester (Reference)a
Women with depressive symptoms in the first or second trimesterb
Adjusted mean scorec (95% CI)
Adjusted regression coefficientsd (95% CI)
62 (34, 91)
−23 (−38, −8)*
−7 (−24, 8)
−24 (−52, 4)
26 (−6, 58)
−10 (−27, 6)
24 (6, 43)*
−20 (−52, 12)
68 (41, 95)
−9 (−23, 4)
2 (−13, 17)
−7 (−33, 20)
71 (55, 88)
−10 (−18, −2)*
18 (8, 27)*
−7 (−24, 9)
65 (44, 86)
−15 (−27, −5)*
27 (15, 39)*
−34 (−55, −13)*
69 (46, 92)
−22 (−34, −10)*
5 (−8, 19)
−50 (−73, −28)*
106 (69, 144)
−41 (−60, −21)*
5 (−17, 26)
−70 (−108, −32)*
85 (71, 99)
−23 (−30, −16)*
13 (5, 21)*
−3 (−17, 11)
Change in Depressive Symptoms and HRQoL in the Third Trimester
Health-related quality of life in the third trimester by change in depressive symptomatology
Women with no depressive symptoms in the second or third trimestera (Reference)
Women with depressive symptoms in the second or third trimesterb
Adjusted mean scorec (95% CI)
Adjusted regression coefficientsd (95% CI)
83 (53, 114)
−31 (−44, −19)*
−3 (−52, −11)*
−15, (−39, 7)
20 (−10, 49)
−28 (−40, −15)*
−13 (−33, 7)
−28 (−50, −5)*
45 (26, 63)
−21 (−29, −13)*
−12 (−25, 0.4)
0.04 (−14, 14)
61 (45, 77)
−12 (−19, −5)*
−7 (−18, 4)
−16 (−29, −4)*
29 (9, 49)
−7 (−15, 1)
0.9 (−13, 15)
−19 (−35, −7)*
67 (44, 90)
−19 (−29, −10)*
−34 (−50, −19)*
−31 (−49, −13)*
53 (11, 95)
−31 (−49, −13)*
−31 (−60, −3)*
−69 (−103, −36)*
85 (74, 97)
−29 (−34, −25)*
−0.8 (−9, 6)
−41 (−50, −32)*
Change in Depressive Symptoms and HRQoL After Delivery
Health-related quality of life after delivery by change in depressive symptomatology
Women with no depressive symptoms in second or third trimester (Reference)a
Women with depressive symptoms in the second or third trimesterb
Adjusted mean scorec (95% CI)
Adjusted regression coefficientsd (95% CI)
81 (43, 119)
4.5 (−11, 19)
−12 (−35, 12)
25 (−0.9, 51)
39 (−15, 93)
2 (−19, 23)
−28 (−62, 5)
−34 (−70, 3)
52 (19, 84)
10 (−2, 23)
8.4 (−12, 28)
−33 (−55, −11)*
87 (57, 116)
6 (−5, 18)
0.03 (−18, 18)
−22 (−42, 2)
43 (16, 69)
−7 (−17, 3)
−15 (−31, 1)
−13 (−31, 5)
56 (23, 89)
6 (−7, 18)
−31 (−51, −10)*
−10 (−32, 12)
74 (29, 120)
−11 (−29, 7)
8 (−19, 36)
−68 (−98, −37)*
92 (78, 106)
7 (2, 13)*
−12 (−20, −3)*
−23 (−32, −13)*
This study expands the current literature on depressive symptomatology and HRQoL by estimating the effect of depressive symptom status on HRQoL during pregnancy and after delivery using a longitudinal approach. We estimated the proportion of women with depressive symptoms during pregnancy and after delivery. Second, we characterized temporal changes in the presence of depressive symptomatology and estimated the effect of the change in depressive symptoms on HRQoL during the pregnancy and after delivery within the context of socio-demographic, clinical and social support factors.
Recent studies highlight the important role of psychological symptoms during pregnancy on maternal anxiety and potentially on newborn outcomes . While the exact mechanisms are not completely understood, early studies suggest a potential deleterious effect of maternal anxiety on early childhood development . O’Connor and colleagues reported a positive association between maternal prenatal anxiety and the incidence of child behavioral and emotional problems at age four . Others suggest that maternal depression may be related to the severity of the initial neonatal illness in premature babies which may have long-term consequences for subsequent growth, neurodevelopment and recurrence of related health problems .
While we had hypothesized that depressive symptoms would be most prevalent after delivery, we found that the proportion of women with depressive symptomatology was greatest in the third trimester (28–32 weeks gestation). This finding suggests that factors related to maternal perceptions of their well-being and surrounding environment may be especially pronounced in the latter stages of pregnancy. Also, it may be that a higher number of pregnancy-related complications, such as gestational hypertension and diabetes that are more likely to develop in the third trimester, contribute to development of maternal depressive symptoms. Because depressive symptoms can be predictive of the development of clinical depression, an interdisciplinary approach by clinicians, psychotherapists and psychologists might help to alleviate current symptoms, and potentially prevent future episodes of clinical depression.
The proportion of women in our sample with depressive symptoms (9–30%) suggests that vigilance is needed throughout the course of pregnancy and after delivery. Rates of depressive symptoms in our study are consistent with findings from earlier investigations. Haas and co-authors , for example, reported similar prevalence rates of 14% and 25% in the first and third trimesters, respectively, and 12% in the postpartum period in a sample largely composed of white and Hispanic women. Our results further contribute to the evidence characterizing the burden of depressive symptomatology during the perinatal period.
We found support for our hypothesis that the presence of depressive symptoms is associated with multiple dimensions of health-related quality of life during pregnancy, even after adjustment for covariates. Women who become depressed and women who remained depressed in early pregnancy (baseline to second trimester) had a substantial decrease in functioning in the second trimester, relative to women with no depressive symptoms. Similarly, women who become depressed and women who remained depressed in the latter stages of pregnancy (second to third trimester) had statistically significantly lower functioning in 7 of 8 functional domains, compared to women without depressive symptoms. Moderately lower functioning was found after delivery. While socio-demographics, clinical conditions and social support account for some of the variation in functioning, the presence of depressive symptoms remained significantly associated with poorer functioning during pregnancy in multiple domains.
The mechanisms underlying the association of depressive symptomatology with functional status is not clearly understood. Further study is needed to determine whether depressive symptoms alter the hormonal milieu during pregnancy, modify neuroendocrine pathways , or create alterations in maternal immune status . The current analysis supports continued surveillance of the prevalence and potential burden of depressive symptomatology on HRQoL, identification of women with depressive symptoms and expedient referrals for clinical intervention.
Our results broaden the knowledge of the effect of a change in depression symptom status on maternal functioning. Our findings suggest that the assessment of depressive symptom status in the first and second trimesters can substantially improve physical and social functioning in mid pregnancy. Health care providers can incorporate screening tools for depressive symptoms into current clinical history questionnaires. Women identified with depressive symptoms can then be referred for early intervention, potential resolution of symptoms and better functioning. Our finding of significantly lower functioning in the third trimester (19–69 points) among women with persistent depressive symptoms suggest that on-going screening of women throughout each trimester may be necessary to lessen the burden of depressive symptoms in reproductive-aged women. Lower scores in Physical Functioning, Social Functioning and Role-Emotional among women who became well in the third trimester suggest that the resolution of depressive symptoms might not directly translate into improvement in physical functioning or social and emotional roles. We found few statistically significant associations between changes in depressive symptoms in the third trimester and HRQoL immediately after delivery. The lack of associations may be due to the timing of the assessments. Women’s perceptions of their HRQoL were assessed within the first 3 days after delivery while they were hospitalized, which is a relatively short postpartum period. Our results may have differed if we had assessed functioning at a later time period after delivery.
Finally, the bi-directional nature of depressive symptoms and HRQoL needs to be explored. It may be that poor functioning precipitates the change in depressive symptomatology. For example, the diagnosis of preterm labor might require reductions in physical functioning. Many women with preterm labor are encouraged to reduce their physical activities, take a leave from work and to go on bed rest. Women may then experience depressive symptoms due to these physical limitations.
Our study has several strengths. First, we describe depressive symptoms and functional status in a racially and socio-economically diverse sample of women in an urban setting. While earlier studies [34, 35] also reported poorer functioning in the perinatal period, few studies have been conducted in a diverse sample of women. Second, we were able to longitudinally assess the effect of depression status on HRQoL. Our study is one of the first to estimate the change in depressive symptom status on physical and social functioning throughout pregnancy and after delivery. Because we used established measures of depressive symptoms and health-related quality of life, we are able to compare our findings in pregnant women with findings from general, non-pregnant patient samples. Average scores for Social Functioning and Mental Health in our study sample, for example, were similar to scores observed in studies of patients with type 2 diabetes mellitus .
There are limitations to our study. First, the study was conducted primarily in women receiving prenatal care within a university-based practice setting. The proportion of women with depressive symptoms and perceptions of functional status might differ if the study were conducted in women from various geographical settings. Second, while our study adjusts for important and relevant clinical factors, it does not include adjustment for physiologic factors (e.g. serotonin, cortisol levels). Third, few women in our sample reported any history of intimate partner violence. We may have overestimated the relation of depressive symptomatology with HRQoL without adjustment for this covariate. The study’s statistical power for race-specific analyses was limited. Alterations in depressive symptomatology during pregnancy may vary with race. Further, the effect of the change in depressive symptomatology on HRQoL may vary with race/ethnicity and cultural perspectives. Also, the sensitivity and specificity of the CES-D may vary slightly in different racial/ethnic groups. However, multiple studies have used the CES-D in multiethnic samples with good validity [18, 37]. Finally, we chose to measure depressive symptoms rather than clinical depression. Sub-clinical depression, however, is a significant clinical problem as evidenced by its effect on health services use and morbidity. Further, non-mental health providers, such as obstetricians, nurse practitioners or nurse midwives, are better able to screen for depressive symptoms in their daily practice than to make a definitive diagnosis of clinical depression.
Our study has important implications. First, our findings underscore the need for health care providers to undergo training to enhance their ability to identify depressive symptoms, and in turn, to make appropriate clinical referrals. Identifying and treating women with depressive symptoms during pregnancy may improve functional status in pregnancy and after delivery. From a clinical perspective, future studies might incorporate physiologic measures in order to better characterize the contribution of hormonal and neurochemical factors to the relation of depressive symptoms with HRQoL in pregnancy and the puerperium. Such studies would lay the foundation for a combination of behavioral and pharmacologic interventions for patients with substantially poor functioning. Finally, studies that prospectively measure the effect of interventions on the change in depressive symptoms status and perceptions of functional status in subsequent pregnancies would provide additional evidence to inform practice guidelines for depression symptom status among child-bearing women.
Dr. Nicholson was funded, in part, by the American Gynecological and Obstetrical Society. Drs. Nicholson and Powe are supported, in part, by the National Institute for Diabetes and Digestive and Kidney Diseases (Grant numbers: 1K23 DK-067944-01 to Dr. Nicholson; 2K24 DKO2643-06A1 to Dr. Powe).